Detection of Linguistic Bad Smells in GRL Models: An NLP Approach

Goal-Oriented Requirements Engineering (GORE) plays a crucial role in facilitating effective communication between stakeholders in system development. Through the use of goal models, GORE provides a structured approach for eliciting, analyzing, and managing requirements from the perspective of stakeholders' goals and intentions. However, goal models are susceptible to poor practices, called also bad smells, that may hinder the effective communication and understanding among stakeholders, potentially leading to misinterpretations and inconsistencies in requirements. In particular, goal models are prone to linguistic bad smells encompassing a spectrum of anoma-lies such as unclear or ambiguous goal statements, conflicting or contradictory requirements, and instances of misspellings. Therefore, identifying and addressing linguistic bad smells in goal models is crucial for ensuring the quality and accuracy of goal models. In this paper, we define seventeen linguistic bad smells in goal models, classified into four categories: Syntax, Semantics, Pragmatics, and Complexity. Furthermore, we provide Natural Language Processing (NLP) based detection methods for twelve specific bad smells, which we have automated to target Textual GRL (TGRL) models. The proposed approach and tool are evaluated using two TGRL models achieving an F2-Score of 0.8.